Key facts
- The U.S. government ordered Anthropic to suspend foreign national access to its Fable 5 and Mythos 5 AI models.
- National security concerns, reportedly related to bypassing safeguards, were cited for the directive.
- Anthropic disabled the models for all users globally to ensure compliance.
- Amazon CEO Andy Jassy raised concerns about Anthropic's AI models to the Trump administration prior to the directive.
The Trump administration has ordered Anthropic to suspend foreign national access to its Fable 5 and Mythos 5 AI models, citing national security concerns. Anthropic confirmed the directive and complied by disabling the models globally for all users, a move that has raised concerns about technological sovereignty.
According to Anthropic, the U.S. government cited national security reasons, reportedly related to a discovered method to bypass or 'jailbreak' the AI models' safeguards. The company stated it received the order without specific details but understood the concern was that a bypass could allow the AI to identify software vulnerabilities. Anthropic disagreed with the decision, noting that the vulnerabilities were minor, present in other publicly available models, and that no universal jailbreak had been found.
To comply with the government's legal order, Anthropic suspended access to Fable 5 and Mythos 5 for all users worldwide. The action marks a significant escalation in U.S. efforts to halt foreign adversaries' AI capabilities, shifting focus from chip export controls to restricting access to AI models themselves.
Anthropic stated that the government has provided only "verbal evidence of a potential narrow, non-universal jailbreak" and disagreed that such a finding should warrant recalling a commercial model deployed to millions. The company believes there has been a misunderstanding and is working to restore access as soon as possible.
The Pentagon's chief information officer, Kirsten Davies, supported prioritizing national security, stating, "Some things are simply more important than revenue cycles, clickbait and pre-IPO valuation. America First. Always."
Experts have noted that models like Mythos could accelerate sophisticated cyber-attacks, particularly in sensitive sectors like banking. Anthropic had previously rolled out Claude Fable 5 with guardrails against risky uses like cybersecurity, which some users found overly broad.
Amazon CEO Andy Jassy was among tech leaders who raised concerns to senior Trump administration officials this week about security risks in Anthropic's most advanced AI models, according to The Information.